amd/blis
1
0
Fork 0
mirror of https://github.com/amd/blis.git synced 2026-04-20 15:48:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

More minor fixes to FAQ.md and Sandboxes.md.

Browse Source
This commit is contained in:
Field G. Van Zee
2021-09-28 18:43:23 -05:00
parent 89aaf00650
commit 3442d4002b
2 changed files with 5 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
docs/FAQ.md
View File

@@ -229,7 +229,7 @@ If this feature is important or useful to your work, we would love to hear from
### Who is involved in the project?
Lots of people! For a full list of those involved, see the
[CREDITS](https://github.com/flame/blis/blob/master/CREDITS) file within the BLIS framework source distribution.
[CREDITS](CREDITS) file within the BLIS framework source distribution.
### Who funded the development of BLIS?
@@ -247,13 +247,13 @@ Feature requests should also be submitted by [opening a new issue](http://github
### I'm a developer and I'd like to study the way matrix multiplication is implemented in BLIS. Where should I start?
Great question! The first thing you should know is that the core framework of [level-3 operations](https://github.com/flame/blis/blob/master/docs/BLISTypedAPI.md#operation-index) was *not* designed to be used to teach or explain a high-performance implementation of matrix multiplication. Rather, it was designed to encode the family of level-3 operations with as little code duplication as possible. Because of this, and also for historical/evolutionary reasons, it can be a little difficult to trace the execution of, say, `gemm` from within the core framework.
Great question! The first thing you should know is that the core framework of [level-3 operations](docs/BLISTypedAPI.md#operation-index) was *not* designed to be used to teach or explain a high-performance implementation of matrix multiplication. Rather, it was designed to encode the family of level-3 operations with as little code duplication as possible. Because of this, and also for historical/evolutionary reasons, it can be a little difficult to trace the execution of, say, `gemm` from within the core framework.
Thankfully, we have an alternative environment in which experts, application developers, and other curious individuals can study BLIS's matrix multiplication implementation. This so-called "sandbox" is a simplified collection of code that strips away much of the framework complexity while also maintaining local definitions for many of the interesting bits. You may find this `gemmlike` sandbox in `sandbox/gemmlike`.
Sandboxes go beyond the scope of this FAQ. For an introduction, please refer to the [Sandboxes](https://github.com/flame/blis/blob/master/docs/Sandboxes.md) document, and/or contact the BLIS developers for more information.
Sandboxes go beyond the scope of this FAQ. For an introduction, please refer to the [Sandboxes](docs/Sandboxes.md) document, and/or contact the BLIS developers for more information.
### Where did you get the photo for the BLIS logo / mascot?
The sleeping ["BLIS cat"](https://github.com/flame/blis/blob/master/README.md) photo was taken by Petar Mitchev and is used with his permission.
The sleeping ["BLIS cat"](README.md) photo was taken by Petar Mitchev and is used with his permission.

2
docs/Sandboxes.md
View File

@@ -121,7 +121,7 @@ Usually, neither of these situations will require any of your local definitions
since those local definitions are only needed to define your sandbox
implementation of `bli_gemmnat()`, and this function is already prototyped by
BLIS. *But if you are adding additional APIs and/or operations to the sandbox
that are unrelated to `bli_gemmnat()`, then you'll want to #include those
that are unrelated to `bli_gemmnat()`, then you'll want to `#include` those
function prototypes from within `bli_sandbox.h`*
5. Your definition of `bli_gemmnat()` should be the **only function you define**